CENTRE FOR SOCIAL SCIENCE RESEARCH

‘Affordability’ and the political economy of social protection in

contemporary Africa

Jeremy Seekings

CSSR Working Paper No. 389

Legislating and Implementing Welfare Policy Reforms

December 2016

Published by the Centre for Social Science Research University of Cape Town

2016

http://www.cssr.uct.ac.za

This Working Paper can be downloaded from:

http://cssr.uct.ac.za/pub/wp/389/

ISBN: 978-1-77011-376-3

© Centre for Social Science Research, UCT, 2016

About the author: Jeremy Seekings is Professor of Political Studies and Sociology, at the University of

Cape Town. Email: jeremy.seekings@uct.ac.za.

Acknowledgements:

An early version of this paper was presented at the UNU-WIDER symposium on ‘The

Political Economy of Social Protection in Developing Countries’, Mexico City,

February 2016. I am grateful for comments from participants at the symposium. This

research was funded in part an award from the UK Department for International

Development through the UK Economic and Social Research Council for the project

on ‘Legislating and Implementing Welfare Policy Reforms’.

1

‘Affordability’ and the political economy of social protection in contemporary Africa

Abstract

The ‘affordability’ of new or expanded social protection programmes depends

on more than an assessment of the fiscal costs or the poverty-reducing or

developmental benefits. Diverse international organisations have showed that

programmes costing less than or about 1 percent of GDP have substantial

benefits, and most low-income countries have the ‘fiscal space’ for such

programmes (including through increased taxation). These international

organisations have generally failed to convince national policy-making elites to

raise and to allocate scarce domestic resources to social protection

programmes. The result is an ‘affordability gap’ between what is advocated for

African countries and what those countries’ governments are willing to spend.

This paper examines four cases of contestation over the ‘affordability’ of social

protection reforms in Africa: Botswana, South Africa, Zambia and the semi-

autonomous territory of Zanzibar. In all four cases political elites resisted or

rejected proposals for expensive reforms. In practice, the most expensive

reforms that were approved were ones costing only 0.4 to 0.5 percent of GDP.

The governments of Zambia and Botswana generally resisted even expenditures

of this magnitude. The cost ceiling for reforms is far below the estimates of

international organisations, reflecting political, normative and ideological

factors.

Introduction

Across much (but not all) of Africa the social protection debate pits international

agencies, aid donors and their local allies (mostly in civil society) favouring the

expansion of social protection – although often in divergent directions – against

domestic political elites who resist proposals that the state should provide direct

income support to the poor. This debate often appears to counterpose empirical

and normative arguments: Advocates of social protection present evidence of the

positive effects of social protection on poverty-reduction and development,

2

whilst local elites assert that states should not let the poor become ‘dependent’

on ‘handouts’.1

The issue of ‘affordability’ spans both empirical and normative approaches.

Whilst often presented as a straightforward empirical issue – what can be

afforded within reasonable fiscal constraints? – the issue of affordability has

inevitable distributional implications that render the issue as much a normative

as an empirical one: Should states impose additional taxes, and on whom, in

order to finance cash transfer programmes? Should they spend scarce resources

on cash transfers, and for whom, if the opportunity cost is undoubtedly

important expenditure on (for example) education or health care or

infrastructure? When local elites resist arguments that social protection

programmes are ‘affordable’, they are almost certainly fusing apparently

technocratic arguments (about the consequences of raising taxes or increasing

debt or depriving other areas of public expenditure) with normative ones (about

who should get what and who should be paying for what).

What states choose to do or not to do is resolved politically. In the 2000s,

neither international agencies nor aid donors have the kind of ‘hard’ power that

they enjoyed in the 1980s, when conditions could easily be attached to bridging

finance. Despite massive aid flows, their power was primarily ‘soft’. This is not

to say that they have no power: Soft power can be potent, not only in terms of

framing the policy-making agenda but also in terms of strengthening reformers

by providing them with the ‘evidence’ that undercuts resistance and facilitating

implementation through technical assistance. Soft power thus shapes and feeds

into political struggles over policy-making, and especially over the raising and

allocation of public revenues.

In this paper, I examine political arguments about, and discourses of,

‘affordability’ in Anglophone East and Southern Africa. The first half of the

paper examines the arguments about affordability put forward by international

organisations including, especially, the ILO. The second half of the paper

examines the politics of affordability in four cases: Botswana (over the fifty

years following independence in 1966), South Africa (focusing on the reform of

child benefits following democratization in 1994, Zambia (focusing specifically

on the mid-2000s, when social protection was resisted strongly) and Zanzibar (in

the mid-2010s, during deliberation over the introduction of universal old-age

pensions). These cases are not presented as being typical or representative.

Rather, they illustrate distinct scenarios in the politics of ‘affordability’. The

1 This juxtaposition is not as neat as it often seems, however: International agencies and aid

donors generally hold a normative position in favour of state interventions (at least for some

groups of deserving poor) whilst the views of local elites are often rooted in the evidence

derived from their personal experience (which they privilege over more systematic studies).

3

four cases include one country (Botswana) that grew from low- to middle-

income, one (South Africa) that was middle-income throughout the period under

review, and two that remained low-income (Zambia and Zanzibar) (see Figure

1).2 The differences in terms of GDP per capita are substantial: By the early

2000s, GDP per capita in Botswana and South Africa was about ten times higher

than in Zambia and Zanzibar.

Figure 1: GDP per capita since 1966

There are also obvious political differences between these four cases. South

Africa (since 1994) and Botswana have governments elected through free and

fair elections, and states with unusually-developed systems of taxation and

social protection. In Zambia and Zanzibar, democracy is clearly flawed, with

persistent allegations that elections have been either unfree or unfair or both, and

their states have less capacity.

Public welfare provision developed along quite different lines in the four cases

(see Figure 2). In Zambia, successive governments have generally been reluctant

to take responsibility for or to expand cash transfer programmes that were

initiated, on an experimental basis, by donors. In Zanzibar, after considerable

debate over affordability, universal old-age pensions were introduced in 2016,

although only for the very elderly and with modest benefits. In South Africa, a

Child Support Grant (CSG) was introduced and repeatedly expanded despite

continuing anxieties about affordability. Botswana, finally, used its rapidly

growing resources from minerals to expand its welfare state, but it did so in

2 Data on South Africa, Botswana and Zambia: World Development Indicators (WDI),

variable NY.GDP.PCAP.KD. Data on Zanzibar collated from various Government of

Zanzibar sources.

0

500

1000

1500

2000

2500

3000

3500

4000

4500

19

66

19

68

19

70

19

72

19

74

19

76

19

78

19

80

19

82

19

84

19

86

19

88

19

90

19

92

19

94

19

96

19

98

20

00

20

02

20

04

20

06

20

08

20

10

20

12

20

14

con

stan

t 2

00

0 U

S$

GDP per capita

South Africa

Botswana

Zambia

Zanzibar

4

distinctly conservative ways, such that its welfare state in the 2000s is quite

different to its similarly-developed neighbour, South Africa. In all four cases

policy-makers worried about affordability, with diverse outcomes.

Middle-income

economy

Low-income

economy

Less generous

welfare

provision

Botswana Zambia

More

generous

welfare

provision

South Africa Zanzibar

Figure 2: Welfare provision and level of development in the 2000s

Estimating the fiscal cost of social protection

International agencies – led by the International Labour Organisation (ILO) and

the World Bank – emphasise that sub-Saharan African countries are, in general,

laggards in terms of expenditure on social protection. The ILO’s World Social

Protection Report (ILO, 2014) and the World Bank’s State of Social Safety Nets

(World Bank, 2015) both present data that suggest that spending in Africa is

consistently lower than in most other parts of the world. The ILO reports, for

example, that Africa has the lowest expenditure in relation to GDP on either

pensions for the elderly or child benefits, and the lowest proportion of

unemployed workers receiving unemployment benefits. Health coverage,

measured in terms of the proportion of the population affiliated to public health

systems or private insurance schemes, is also lowest in Africa, at 25 percent of

the population compared with 61 percent globally (ILO, 2014). In Africa, only

2.8 percent of GDP is spent on ‘social security’, compared with a global average

of 5.8 percent, according to World Bank studies (World Bank, 2012b, 2015; see

also Garcia and Moore, 2012).

Elsewhere I have argued that welfare regimes in Africa should be viewed as

moving down a distinct trajectory rather than as laggards following countries in

other regions down a standard path (Seekings, 2013, 2014). This is in part

because various categories of expenditure – including especially agrarian

5

welfare programmes3 and food programmes4, as well as programmes funded by

non-state agencies – are often excluded from the data. Regardless of the merits

of this argument, it is clear that there is significant international pressure on

African governments to expand their apparent expenditures on social protection,

and international agencies and aid donors emphasise strongly that expanded

expenditure is affordable. African governments themselves have nominally

committed themselves to the expansion of social protection, although with a

strong emphasis on strengthening the family. Every two years since 2008, the

African Union has convened a conference of ‘Ministers in charge of Social

Development’. The first conference, in Windhoek (Namibia), approved a Social

Policy Framework for Africa and adopted the Windhoek Declaration on Social

Development.5 The AU also endorsed social protection in its Ouagadougou

Declaration (2004), and Livingstone Declaration (2006) (Wright & Noble,

2010).6

What it would cost to expand social protection depends, of course, on how

extensive the system would be. Evidence on the costs of programmes comes

from two sources: The actual costs of existing programmes, including pilot or

experimental programmes that can be scaled up; and modeling the prospective

costs of possible programmes, using appropriate economic and demographic

data to gauge the effects of conditions on coverage. Given that administrative

costs and other overheads tend to be low (except for public employment

programmes, where supervision and materials can account for a significant

fraction of total costs), the most important determinants of costs are the

eligibility conditions (such as the age threshold for pensions and any means-test

for poverty-targeting) and the generosity of the benefits provided.

3 For example, should programmes such as Malawi’s Agricultural Input Subsidy Programme

(AISP), later transformed into the Farm Input Subsidy Programme (FISP), be counted as

social protection (Hagen-Zanker and McCord, 2010)? 4 The World Bank’s State of Social Safety Nets 2015 includes (I think for the first time) in-

kind transfers and school feeding programmes in its assessment of social safety nets. It reports

that, globally, 718 million people received cash transfers, including through workfare, whilst

600 million received in-kind transfers, 276 million participated in school feeding

programmes, and 381 million benefitted from fee waivers or targeted transfers. 5 Subsequent ministerial conferences were held in Khartoum in November 2010, Addis Ababa

in November 2012 and again in May 2014. 6 Wright & Noble note that the reference to social protection in the Framework was inserted at

the last minute. Most poverty reduction strategy papers (PRSPs) and national development

plans (NDPs) initiated between 2005 and 2010 mentioned social protection, but most did not

include any substantial discussion of it (UNECA, 2011).

6

Modelling the costs of prospective programmes, 2005-10

In 2005, both the ILO and World Bank completed their first detailed studies

costing social protection in lower-income African countries. Kakwani &

Subbarao (2005, 2007), for the World Bank, examined the role of pensions in

reducing poverty among the elderly in 15 African countries. They concluded

their study with a simple cost calculation: A universal social pension, set at 70

percent of each country’s poverty line for men and women from the age of 60

would cost between 1 and 4 percent of GDP. The cost would decline massively,

however, if benefits were more parsimonious (set at, say, 35 percent of the

poverty threshold), the age of eligibility was higher (65 rather than 60 years), or

the programme was targeted on the poor (through a means-test). Kakwani &

Subbarao concluded that the case for a generous, universal pension was weak in

terms of both reducing poverty and fiscal affordability. Allocating substantial

sums to pensions would clearly crowd out other, more important programmes,

including basic health care or provision for poorer groups of people including

children. Kakwani & Subbarao were more positive, however, about a targeted,

parsimonious pension.

At the same time an ILO team costed a set of social assistance programmes for

seven low-income countries in Africa (Burkina Faso, Cameroon, Ethiopia,

Guinea, Kenya, Senegal, and Tanzania).7 Pal et al. (2005) costed universal old-

age pensions, child benefits and disability grants, with different benefit levels

and (for children) eligibility conditions. A universal old-age pension of US$

0.50/day (adjusted for purchasing power) paid to all men and women above the

age of 65 would cost about 0.5 percent of GDP. The cost would be larger if

benefits were set at a higher level, such as (in the less-poor countries) 30 percent

of GDP per capita. Alternative child benefit programmes were costed at between

1 and 4.5 percent of GDP, with the cheapest programmes covering only double

orphans. Pal et al. also costed a targeted grant to the poorest 10 percent of

households, with benefits of just under US$14/household/month (modelled on

an experimental cash transfer programme in Zambia). The cost of this would

vary from 0.15 percent to a high of 0.7 percent of GDP (with the proportional

cost being highest in the poorest countries, i.e. Ethiopia and Tanzania).8

7 These programmes were costed as part of a larger social protection package that included

also basic health care and education, which were very much more expensive. Pal et al. also

costed administrative overheads at 15 percent of benefits. A separate study costed similar

programmes in five South Asian countries. A further study was envisaged of selected

countries in Latin America. 8 Pal et al. costed administrative overheads at 15 percent of benefits. Pal et al. also costed

basic health care and education, which would require very much larger expenditures.

7

The ILO’s costing for the African cases was updated and extended in 2008

(ILO, 2008a).9 The updated analysis costed the following programmes (together

with administrative costs):

1. An old-age pension programme, set at 30 percent of GDP per capita

(on the basis of a Tanzanian study of the cost of living), with a maximum

of US$ 1/day (adjusted for purchasing power), paid to all men and women

above the age of 65; together with a disability pension programme paying

the same benefits to 1 percent of the working age population; the total

cost varied between 0.6 percent and 1.5 percent of GDP.

2. Basic child benefits, set at 15 percent of GDP per capita, with a

maximum of US$ 0.50/day (PPP), payable for each of two children to the

age of fourteen, to the mother; the total cost varied between 1.5 percent

and 3.5 percent of GDP.

3. Social assistance to un- or underemployed working-age adults

through workfare, along the lines of the Indian National Rural

Employment Guarantee Scheme, that guaranteed low-wage work for 100

days p.a. per household; the ILO assumed that the programme would need

to reach 10 percent of the working-age population (excluding households

benefitting from either old-age pensions or child benefits); benefits would

be set at 30 percent of GDP per capita, with a maximum of US$1/day

(PPP); the cost would vary between 0.3 percent and 0.8 percent of GDP.

The ILO team also costed basic, universal health care.

Soon after these, a third international agency – UNICEF, in conjunction with the

Overseas Development Institute (ODI) in the UK – conducted a set of studies of

the costs of (and possible financing options for) child benefits and old-age

pensions in five west and central African cases (Congo, Equatorial Guinea,

Ghana, Mali, Senegal) (Handley, 2009).10 Three cash transfer programmes were

costed:

A universal child benefit: Payable for all children to the age of 14,

with benefits set at 30 percent of the extreme (food) poverty line; the

cost would be a modest 0.9 percent and 2 percent of GDP in oil-rich

Equatorial Guinea and Congo respectively, but would be between

about 6 percent of GDP in each of Mali and Senegal, and a massive 9

percent of GDP in Ghana.

A targeted child benefit: Payable for all children to the age of 14,

with benefits set at 30 percent of the extreme (food) poverty line (as

9 The authors were Behrendt and Hagemeyer. 10 The costing studies were undertaken by various authors: Notten et al on Congo, Barrientos

and Bossavie on Mali and Senegal, Barrientos on Equatorial Guinea and Ghana.

8

with the universal child benefit), but only for children in households

below the poverty line; this would cost just over one half of the

universal child benefit option.

An old-age pension programme: Payable to men and women above

the age of 60, with benefits set at 70 percent of the poverty line; in the

Equatorial Guinea this would cost a paltry 0.2 percent of GDP, and in

the Congo it would cost only 1 percent of GDP; in Ghana, it would

cost 2.6 percent of GDP (and it was not costed in either Mali or

Senegal).

Administrative costs were assumed to be 10 percent for the universal

programmes and 15 percent for the targeted one. The UNICEF study discussed

in some detail the fiscal space in each country for these kinds of programmes.

These UNICEF costs were very much higher than the ILO’s. For Senegal –

which was the only country in both studies – UNICEF costed the universal child

benefit at 6.4 percent of GDP, and the means-tested child benefit at 3.7 percent

of GDP, whereas the ILO costed the universal child benefit component of the

‘basic social protection package’ at 2.3 percent of GDP in 2005 (dropping to 1

percent over time) (Pal et al., 2005: 24). The reason for this presumably lay in

the level of the benefit: 30 percent of the national poverty line (as used by

UNICEF) must have been much more generous than the US$0.25/day (PPP)

(used in the ILO studies).

Since the mid-2000s there have been numerous studies of the costs of cash

transfer programmes in individual countries. In Senegal, for example, a child

benefit programme was costed at 1.7 percent of GDP by Samson & Cherrier

(2009, cited by Schnitzer, 2011). Schnitzer (2011), however, assessed that

Samson & Cherrier’s proposals were “likely not to be affordable in the current

situation of recovery from the economic crisis and likely fiscal readjustment and

budget austerity in the few years to come” (2011: 18). A more modest

programme was more realistic, and could be funded through reallocating

resources spent on food and fuel subsidies, which cost 3 percent of GDP (and

benefitted primarily the non-poor). Schnitzer costed a child grant programme

limited to the fifteen poorest districts at 0.55 percent of GDP. In Tanzania, to

take another example, ILO researchers costed universal pensions in 2006 and

2008, going into much more detail than in the multi-country studies conducted

in 2005 and 2008 (ILO, 2006, 2008c). In 2010, both the UK-based NGO

HelpAge Internatonal and the local (Tanzanian) NGO REPOA costed universal

pensions, examining the cost implications of different age thresholds and benefit

levels. HelpAge (2010) costed different versions at between 0.26 and 1.28

percent of GDP, whilst REPOA (2010) costed their versions at between 1.1 and

2.3 percent.

9

The costs of existing programmes

As more programmes were initiated across Africa, more data slowly became

available on actual costs. O’Cleirigh (from Irish Aid, in a report for the OECD)

referred to the PSNP in Ethiopia, which targeted the 10 percent of households

that were most food-insecure in the long-term, at a cost of 1.7 percent of GDP.

O’Cleirigh also pointed to the experience of a five-month long district-specific

feeding programme in Malawi. Scaling up the programme to cover the 6 percent

of households experiencing food insecurity across Malawi as a whole would

cost 0.5 percent of GDP. Expanding the programme to the poorest 10 percent of

households for the entire year would cost 1.8 percent of GDP (O’Cleirigh,

2009). Garcia & Moore (2012: 175-8, for the World Bank) collated data on

more than one hundred cash transfer programmes across Africa. Most were still

in pilot stages, and the costs were generally negligible. The larger programmes –

such as the PSNP in Ethiopia, and universal old-age pension programmes in

Namibia and Lesotho (as well as in South Africa) – cost more than 1 percent of

GDP. Programmes that paid more modest benefits – such as old-age pensions in

Botswana – generally cost less (although the South African child grant, payable

to most children to the age of eighteen, cost more than 1 percent of GDP).

The experience of actual programmes revealed the complexities of cost. The

World Bank tends to favour targeted programmes in part on the basis that they

are more affordable. The ILO favours less generous universal programmes. In

the mid-2000s, however, South Africa’s means-tested old-age pension scheme

cost just over 1.2 percent of GDP,11 whereas Botswana’s universal12 pension

cost about one-fifth of this in relation to GDP.13 These cases remind us that costs

depend on benefit levels as well as eligibility conditions: South Africa’s pension

was generous whereas Botswana’s was parsimonious.14

Fiscal space

From the outset, even the ILO recognized that anything approaching a ‘basic

social protection package’ would be beyond the means of most low-income

countries on their own, although the reallocation of domestic resources would go

11 As we shall see below, the old-age pension was then paid to women from the age of 60 but

to men from the age of 65. Later, the age of eligibility for men was lowered to remove this

discrimination. The cost of the programme rose marginally. 12 From age 65. 13 PensionWatch report that, in 2010, the cost was 0.26% of GDP. 14 In 2010 the pension benefit was US$56 per month in purchasing power. By contrast, the

pension in South Africa was worth US$248 per month in purchasing power. See

PensionWatch database.

10

a long way). Pal et al.’s calculations showed that external financing would need

to account for the lion’s share of expenditure on their most ambitious proposals.

The 2008 ILO report reiterated this:

‘The projections show that introducing a complete package of basic

social security benefits requires a level of resources that is higher than

current spending in the majority of low-income countries (which

rarely spend more than 3 per cent of GDP on health care and rarely

more than 1 per cent of GDP on non-health social security measures).

Therefore, a considerable joint domestic and international effort is

needed to invest in basic social protection to bring about significant

social development and a sharp reduction of poverty’ (ILO, 2008a:

11).

Pal et al. (2005: xii) emphasized, however, that “a basic social protection benefit

package can be affordable if it is made a priority area of national policy”, i.e. if

governments reallocated expenditure to social protection. “If the national

commitment exists and one third of total government expenditure can be

reallocated to meet basic social protection needs then the necessity for

international financing would show a steady decline in the medium-term” (2005:

41). The 2008 ILO study pointed to the possibilities of funding health care in

part through a contributory programme (along the lines of Ghana’s National

Health Insurance system) and in part through debt financing, with donor support

to fill any ensuing gaps. The ILO concluded that they had shown that “a basic

social protection package is demonstrably affordable”, but

‘on condition that the package is implemented through the joint efforts

of the low-income countries themselves (reallocating existing

resources and raising new resources, i.e. through health insurance or

other earmarked sources of financing for social security) and of the

international donor community – which would in some cases have to

refocus international grants on the supplementary direct financing of

social protection benefits, on strengthening the administrative and

delivery capacity of national social protection institutions in low-

income countries and on providing the necessary technical advice and

other support’ (ILO, 2008a: 18).

In 2006, the ILO calculated that it would cost less than 2 percent of global GDP

to provide basic benefits to all of the world’s poor (cited in ILO, 2008a: 3).

Other studies were less sanguine than the ILO. The UNICEF study showed that

some countries – notably oil-producing Congo and Equatorial Guinea – had

substantial available resources, although their governments chose to spend very

11

little on social protection. ‘Affordability’ was ‘more of a problem’ in the other,

‘aid-dependent’ countries (Handley, 2009). O’Cleirigh noted that both domestic

tax revenues and donor aid were rising, and concluded that “the levels and

trends of revenue growth would seem to imply that financial affordability should

not be a binding constraint to financing modest but significant social protection

programmes” (2009: 121, emphasis added).

The clearest warning came from Hagen-Zanker & McCord (2010), in a report

for the ODI in the UK. They noted that African governments would need to

balance spending on social protection (narrowly-defined in terms of income

support) with spending on other pressing priorities. Most African governments

had formally endorsed a set of spending targets covering health, education,

sanitation, agriculture and rural development, and infrastructure, in addition to

social protection (as shown in Hagen-Zanker & McCord’s Table 1, below; from

2010: 9). If governments are spending below other targets, then social protection

will be competing with other public policy areas for additional funding.

Hagen-Zanker & McCord examined what these would mean for five African

countries (Ethiopia, Kenya, Malawi, Mozambique and Uganda). Using data

from 2006-07 (which predated some of the spending targets listed above),

Hagen-Zanker & McCord found that three of their five countries’ governments

met the targets with respect to education spending, only two did so with respect

to agriculture and only one did so with respect to health. None came close to the

targets for social protection, water and sanitation or infrastructure (see their

Table 3, p13). Hagen-Zanker & McCord extended their analysis to include ‘off-

budget’ official development assistance (ODA) – i.e. ODA not recorded in data

on government expenditure15 – for two of their cases (Malawi and Uganda). In

Malawi, off-budget donor aid pushed aggregate spending on health and

agriculture even further above the target, and raised spending on education and

water/sanitation close to the target. Social protection remained far below the

15 Government expenditure including government-recorded ODA came to between 20 and 27

percent of GDP in their five cases. I estimate (on the basis of the data reported by Hagen-

Zanker and McCord) that ‘off-budget’ ODA constituted as much as another 10 percent of

GDP in Uganda and perhaps even more in Malawi.

12

target, as did infrastructure. In Uganda, off-budget ODA made little difference,

and even aggregate spending remained below-target in all six categories.

Certainly in Uganda and probably in Malawi, social protection faced stiff

competition from other sectors for resources. The intensity of this competition is

evident from Hagen-Zanker & McCord’s calculations of how big an increase in

spending would be required to meet all of the targets. Even if all government

expenditure was reallocated to these six expenditure categories, only one of their

five countries (Kenya) could meet the targets (2010: 18).

The World Bank’s response has generally been to emphasise the possibilities of

reallocating existing public expenditure as well as harnessing the resources

becoming available because of economic growth. In 2012, the World Bank

pointed to low levels of expenditure on social protection across most (but not

all) of Africa, but insisted that “achieving national coverage” was “fiscally

affordable”. Economic growth was expanding the ‘fiscal space’, domestic

resources could be shifted from other programmes – especially from poorly-

targeted (and often distortionary) general price subsidies (which, in the case of

Senegal, cost 3-4 percent of GDP in the late 2000s), and donor funding could

continue to play an important role (World Bank 2012a, 2012b).

As the World Bank itself acknowledged, however, there was a clear political

challenge:

‘Increasing policymakers’ awareness of the links between social

protection and economic growth will be imperative for building

political coalitions in support of social protection funding, by

overcoming concerns that social protection promotes dependency. In

other countries, a case should be made for increasing financing for

social protection as a means of realizing the constitutional rights of

citizens, such as in Kenya and South Africa’ (World Bank, 2012a: 4).

The World Bank recognized that policymakers in low-income countries in

Africa and elsewhere would not prioritise (or support at all) social protection if

they saw transfers as ‘handouts’. This pointed to the imperative of countering

such perceptions with evidence of the developmental benefits of social

protection. The World Bank also recognized that, politically, more spending was

not always productive: Social protection expenditure “should be focused on

efficient and effective programs that have been evaluated, proven to be

effective, and then scaled up”, with programme benefits set at ‘affordable’ levels

and well-targeted (2012b: 65).

13

Revisiting affordability in the 2010s

The international agencies’ approach to the affordability of social protection

shifted from 2009, in response to two main factors. First, the global economic

crisis pushed diverse organisations to embrace more fully a ‘social protection

agenda’. Various UN agencies agreed to the ILO’s proposal to launch a Social

Protection Floor Initiative as one of nine UN joint initiatives to cope with the

effects of the economic crisis. In 2010, the ILO published a report on Extending

Social Security to All: A guide through challenges and options (as well as a

once-off World Social Security Report). The Social Protection Floor Advisory

Group (chaired by former Chilean president, Michele Bachelet) recommended in

2011 that countries adopt social protection floors, although the Group

acknowledged that there was no template that would suit all contexts. The

following year, the ILO adopted Recommendation #202 on Social Protection

Floors. The World Bank also restated its case for social protection – including in

Africa (see World Bank, 2012a, 2012b). Whilst the ILO and World Bank

differed in their priorities, both endorsed the expansion of social protection.

Both the ILO and the World Bank renewed their efforts to demonstrate that

social protection was affordable.16 As part of its preparations for its 2014/15

World Social Protection Report, the ILO costed a universal child benefit

programme in more than fifty low- and middle-income countries using

standardized criteria. The ILO costed a programme paying benefits set at 12

percent of the national poverty line (this being the average ratio in Europe) for

non-orphaned children up to age 18, together with more generous benefits for

double orphans (set at the poverty line) and modest administrative costs. The

average cost would be 1.9 percent of GDP (using an arithmetic average for 57

countries, or 0.9 percent using a weighted average). The cost was higher in

countries with higher poverty, a higher poverty line or more children in the

population. In one-third of their overall set of countries, the cost would be less

than 1 percent of GDP. This sub-set included only two African countries,

however (Cape Verde and Ghana). The projected cost exceeded 3 percent for

only one in five countries in the global set, but these included many African

countries, especially in West Africa, and also Kenya, Ethiopia, Malawi and

Mozambique. These costings did not adjust for savings on existing programmes,

but these were unlikely to be significant (ILO, 2015a). The World Bank began

to publish its own annual review of “the state of social safety nets” in 2014, with

the explicit goal of disseminating data and analysis to guide policy-making. In

its 2015 State of Social Safety Nets, the World Bank proclaimed boldly and

16 On 30 June 2015 the World Bank and ILO put out a joint statement ‘calling the attention of

world leaders to the importance of universal social protection policies and financing’.

14

unambiguously that “safety nets are affordable at all levels of income” (World

Bank, 2015: 21).

The second shift in the 2010s was a growing focus on revenues as much as

costs. A series of studies considered fiscal space in selected countries. For

example, Handley (2009) examined fiscal space in five West and Central

African countries, showing that some (especially the oil-rich ones) clearly had

ample fiscal space for new social programmes, but some other countries faced

tougher challenges. In another example, Aguzzoni (2011) examined fiscal space

in Zambia for the ILO, three years after the ILO’s detailed costing of social

protection programmes there. Gough & Abu Sharkh (2011) conducted cluster

analysis on the composition of public revenues for a set of 59 developing

countries, distinguishing between clusters on the basis of whether countries

relied primarily on domestic taxation, social security contributions, aid or other

revenues, or had no dominant source of revenue. Most of their African cases

were either tax-dependent or aid-dependent.

In 2015, the ILO’s Ortiz et al. collated evidence on fiscal space in a large

number of middle- and low-income countries. “The argument that spending on

social protection is unaffordable is becoming less common in international

development forums”, they stated at the outset: “Finding fiscal space for critical

economic and social investments is necessary for inclusive growth as well as for

sustained human development, particularly during downtimes” (ILO, 2015b: 1).

Ortiz et al. discussed eight mechansisms for expanding the fiscal space for

social protection. These included: the re-allocation of public expenditures from

“areas with limited development returns”, for example from military expenditure

to health (as in Costa Rica and Thailand); increasing tax revenues, as in Brazil

(which had introduced a financial transactions tax) and Bolivia and Mongolia

(which used mining and gas revenues); expanding social security coverage and

contributory revenues, as in much of Latin America; and lobbying for additional

aid and transfers.17 Additional mechanisms included the elimination of illicit

financial flows, using fiscal and foreign exchange reserves, borrowing or

restructuring existing debt and adopting a more accommodative macroeconomic

framework. Ortiz et al. collated basic data on fiscal space for middle and low-

income countries, revealing whether each country was collecting little tax or

spending a lot on the military. Unfortunately, key fiscal data were unavailable

for a number of countries. Their conclusion was clear: Even the poorest

countries had the fiscal space to expand social protection for the poor.

17 O’Cleirigh (2009) noted that the cost of universal social protection was a small fraction of

the total aid flowing into many countries.

15

Analyses of fiscal space suggested that, in the words of the then Director-

General of the ILO (Juan Somavia), “the world does not lack the resources to

abolish poverty, it only lacks the right priorities”.18 Taxation is, of course, as

political as public expenditure, although often in different ways. The growing

attention paid to taxes and fiscal space generally prompted pioneering studies of

the political economy of taxation, for example by UNRISD with respect to

Uganda (including Kjaer & Ulriksen, 2014).

Case-studies of the political economy of affordability in Africa

The second half of this paper examines four case-studies of the political

economy of affordability in Africa. In each case, there appears to be an

‘affordability gap’ between international agencies and aid donors, advocating

expanded social protection and claiming that there was fiscal space, and

domestic political elites who expressed anxiety about ‘affordability’, perhaps

because they did not really buy into the whole idea of social protection. The first

case-study is Botswana, where for several decades the political elite expressed

ambivalence about social protection at the same time as they gradually expanded

it. The second case-study is of South Africa, where the African National

Congress-led government inherited in 1994 an extensive welfare state, but then

had to decide whether and how to reform it. The final two case-studies are of

low-income countries. In Zambia, the government resisted pressure to expand

social protection in the mid-2000s, before a more populist government agreed to

modest reforms after 2011. In Zanzibar, in 2016, the government introduced

universal old-age pensions, albeit only from the age of 70 and with modest

benefits, despite severe fiscal constraints.

Botswana

Botswana has experienced exceptional economic growth since independence in

1966, lifting the country from being one of the poorest countries in the world to

a middle-income economy. By 2008, GDP per capita was fourteen times higher

in real terms than it had been at independence, and was much the same as in its

affluent neighbor, South Africa. Growth was fueled primarily by the rapid

expansion of mining, from the early 1970s. Mining fueled also the rapid growth

of government revenue and public expenditure, as a result of the government’s

shrewd and public-spirited negotiations with mining companies. Economic

18 Social Protection Floor website.

16

growth made possible the expansion and institutionalization of a welfare state.

The fact that the benefits of growth were not shared across the whole population

also provided strong impetus towards remedial state interventions. Through the

first five decades following independence, however, government ministers

repeatedly spoke out against ‘dependency’ on ‘handouts’. The government

regularly expressed concern over the affordability of its welfare programmes,

the cost of which (less than 2 percent of GDP in the 2010s) is much less than in

South Africa or Mauritius (although it is more than in most lower-income

African countries). The construction of a welfare state in Botswana is a story of

fiscal and programmatic expansion despite the generally conservative attitude of

its governments, which explains why the welfare state has distinctly

conservative features (Seekings, 2016c).

The fiscal expansion of the welfare state in Botswana entailed a series of phases,

reflecting in part the fiscal space as perceived by the government. The first

phase entailed large-scale drought relief and then post-drought recovery

programmes in the mid- and late 1960s. The Bechuanaland (later Botswana)

Democratic Party (BDP) won elections in early 1965 and proceeded to form a

government, with Seretse Khama as Prime Minister. At independence, in

September 1966, Khama became president of the new Republic of Botswana.

This took place in the middle of the worst drought in many decades. Faced with

the prospect of famine and the devastation of cattle herds, the newly-elected

government was compelled to provide drought relief – even though it had no

money. At independence, grants from the British government funded more than

one half of the government of Botswana’s recurrent expenditure and all of its

developmental expenditure. Expenditure was ‘acutely constrained by a shortage

of funds’ (Colclough & McCarthy, 1980: 85). The drought relief and recovery

programme was made possible – and its design shaped – by the availability of

large volumes of free food (and even stockfeed) through the new World Food

Programme (WFP). Between mid-1965 and mid-1969 the WFP spent

approximately $15 million on Botswana, which was about one half of the total

domestic revenues (i.e. excluding grants from the UK) of the government of

Bechuanaland/Botswana over this period.19 The cost of the drought relief and

recovery operation ran at about 2 percent of GDP at this time (Stevens, 1978),

but was nonetheless far beyond the current fiscal capacity of the government of

Botswana. The fact that almost all WFP aid was provided in kind – as food or

stockfeed – reinforced the preference for in kind benefits within the new

political elite, with lasting implications for the design of the welfare state in

Botswana.

19 The total WFP expenditure over 1965-69 is calculated from data provided by the WFP (see

Seekings, 2016a: Appendix 1). The exchange rate at the time was about R0.7:$1 (according to

World Development Indicators). The domestic revenues of the government of Botswana were

about R6-7m p.a. at this time (Colclough and McCarthy, 1980: 80).

17

The second phase of welfare state-building in Botswana was based on further

drought relief and recovery programmes over the following twenty-five years.

Drought recurred in 1979-80, through most of the mid-1980s, and again in the

early 1990s. In contrast to the 1960s, however, the government of Botswana

enjoyed rapidly growing domestic revenues. In the late 1960s the government’s

fiscal priority was to raise additional revenues to free it from dependence on

grants provided by the UK (and hence the necessity of securing British approval

of all additional expenditures). Expatriate technocrats helped to renegotiate the

terms of the Southern African Customs Union, resulting in a large increase in

customs revenues; income tax revenues also grew rapidly. Domestic revenues

rose fourfold (in current prices) between 1968/69 and 1972/73, by when the

recurrent budget was balanced and did not require British subvention (Colclough

& McCarthy, 1980). Despite improved public finances and economic growth,

the WFP continued to provide substantial food aid, through both regular feeding

programmes (especially in schools) and emergency relief during and after

drought.

Although a large part of the total cost of welfare provision – between 1 and 3

percent of GDP – continued to be borne by the WFP, the government of

Botswana did have to make occasional decisions about their financial

commitments. The government assumed almost all the non-food costs of relief

and recovery programmes. In 1984, during chronic drought, Botswana received

about 31 million Pula in food aid (mostly from the WFP) and actually spent

about P27 million, almost of which came from domestic revenues (Holm &

Morgan, 1985: 476). The total cost came to about 3 percent of GDP, divided

approximately equally between donors and the government of Botswana.

Outside of droughts, the government incurred costs in running feeding

programmes (although the food itself came from abroad) and for ‘destitute’

relief, but the latter at least was very inexpensive. Destitute relief served, like the

poor laws in nineteenth century Europe, to mitigate extreme poverty, but the

government looked primarily to economic growth and agricultural programmes

to reduce poverty.

In the second half of the 1990s, the subsidization of the welfare state through the

WFP (as well as development aid from the UK and elsewhere) largely dried up.

Some foreign aid continued: The US President’s Emergency Plan for AIDS

Relief (PEPFAR) funded Botswana’s Orphan and Child Programme until 2013.

In general, however, it was during this period that the government of Botswana

assumed financial responsibility for programmes hitherto funded externally

(especially the feeding programmes) and introduced new programmes (such as

the old age pension) even though the cost was borne entirely by the government

itself. Both of these reforms entailed significant additional financial

commitments. Firstly, the new old-age pensions cost almost P100 million, or 0.5

18

percent of GDP, in 1997-98. The cost in relation to GDP declined subsequently,

to about 0.2 percent, because the real value of the pension was not increased

whilst the GDP continued to grow. It is likely that the government anticipated

this when it decided to introduce the pensions. Secondly, the government

assumed full responsibility for the feeding programmes. Outside of drought (and

drought recovery) years, the cost of feeding programmes was modest. In

1997/98, the government spent only P42 million or 0.2 percent of GDP,

although expenditure rose subsequently to about 0.5 percent of GDP during the

drought of the early 2000s. Although the government’s detailed financial

projections do not seem to be public, the government probably anticipated that

these two reforms would cost it close to 1 percent of GDP. This level of

expenditure would have been a massive outlay for many African countries, but

for Botswana it was less than was being spent on drought relief and recovery

programmes for much of the time. It was also deemed affordable given the

apparently healthy condition of public finances. The expansion of mining had,

thanks to a shrewdly-negotiated deal with De Beers, resulted in a massive

increase in government revenues from the 1970s. At the time that the

government was considering the old-age pension and the feeding programmes, it

was projecting continued budget surpluses. The 8th National Development Plan

(NDP, for the years 1997/98-2002/03) forecast that rising revenues would

continue to outpace rising expenditure (although the NDP did warn that a

combination of slower-than-forecast economic growth and unrestrained public

expenditure would result in budget deficits, and hence slower growth –

Botswana, 1997a: 69-71). Quett Masire, who served as Khama’s vice-president

and Minister of Finance and Development Planning, had been said to be “thrifty

to the point of being miserly”, according to fellow-BDP leader David Magang;

after Masire succeeded Khama as president (in 1980), “there dawned the era of

freebies, which included literally free, throwaway money” (Magang, 2008: 475).

In the late 1990s, as Masire handed over the presidency to his deputy, Festus

Mogae, the budget slid into unanticipated crisis. Economic growth was better

than expected but public expenditure rose so fast that the government incurred a

massive budget deficit of P1.4 billion or more than 6 percent of GDP in

1998/99. This was the first budget deficit since 1982/83 (when the deficit had

been about a modest 2 percent of GDP). Deficits recurred also in 2001/02 and

2002/03, as expenditure continued to rise whilst revenues in fact declined

(Botswana, 2003: 35-39). Increased domestic funding of welfare programmes in

the late 1990s contributed to these budget deficits, although rising expenditure

was driven primarily by the public sector wage bill. In response, the government

introduced VAT in July 2002, and proposed ‘cost recovery’ for public services.

This fiscal crisis formed the backdrop for the fourth phase of welfare state-

building in Botswana, in the early 2000s. The government agreed to a series of

19

mostly parametric reforms that raised the total costs of its social assistance

programmes to close to 2 percent of GDP despite economic growth. The value

of food baskets was raised, and workfare was expanded. But the government

rejected proposals for two expensive reforms: for a Child Support Grant which

would have cost 1.2 percent of GDP, in 2010, and for a Family Support Grant

costing up to 0.35 percent of GDP, three years later. This was a period of

renewed anxiety over the cost of welfare programmes. Public finances had

recovered in the early 2000s after the unprecedented budget deficit of 1998/99.

The global economic crisis of 2008-09 caused commodity process to crash,

however, which meant that the government of Botswana’s revenues from

mining collapsed. The 10th NDP (2009-16) budgeted for deficits, and warned

that it would be difficult to maintain safety nets (Botswana, 2009: 46). The need

for fiscal caution was restated in 2013 in the mid-term review of the 10th NDP

(Botswana, 2013b).20

Fiscal considerations were clearly important in this process of policy reform.

Large-scale drought relief and recovery programmes could be adopted at the

time of independence in large part because most of the cost was shouldered by

the WFP. Improved public finances (together with continued support from the

WFP through to the 1990s) enabled the subsequent institutionalization of these

programmes, eventually at the expense of the government of Botswana alone,

and their expansion through the old-age pension and other programmes. The

fiscal crisis of 1998-2002 and the later global economic crisis from 2008 both

heightened fiscal anxiety, prompting government to reject proposals for

expensive reforms.

Botswana’s welfare programmes evolved very consistently, despite these

changes over time in their perceived affordability. In the late 1960s and early

1970s, Khama, Masire and the BDP developed a welfare doctrine that justified

the state’s provision of relief, primarily in the face of drought but also taking

into account the erosion of ‘traditional’ norms of extra-state support by kin or

within the ‘community’, with benefits at a minimal level, often in kind rather

than cash, and generally to families rather than individuals. This doctrine

reflected political conditions, with the elected government assuming roles

previously played by hereditary chiefs and the BDP securing its support base in

rural areas against the threat posed by competing political parties (Seekings,

2016b). Subsequent reforms to the welfare state were generally incremental. The

most dramatic reform – the introduction of old-age pensions – represented the

modernization of the existing and insufficient system of relief for ‘destitutes’.

The government’s over-optimistic assessment of public finances allowed it to 20 Isaac Chinyoka tells me that the anxiety over the sustainability of social assistance is rooted

in the fear that diamonds are not forever. This point is also made by President Ian Khama in

an interview by Greg Mills, Daily Maverick, 29th June 2016.

20

introduce a programme that would cost about 0.5 percent of GDP, but the cost

was contained by setting benefits at a level that was very parsimonious relative

to neighbouring South Africa (where the old-age pension programme therefore

cost three times as much in relation to GDP). The other similarly ‘expensive’

reform was the government’s assumption of responsibility for the school and

other feeding programmes hitherto supported by the WFP. The government

rejected more expensive reform proposals.

The Child Support Grant in South Africa21

When South Africa democratized in the early 1990s it was already a middle-

income economy with an extensive set of cash transfer programmes, although

the apartheid-era National Party government had repeatedly insisted that the

country did not have a ‘welfare state’. The decline of subsistence agriculture in

South Africa meant that drought ceased to be a significant cause of poverty.

From the 1930s and 1940s – i.e. almost half a century earlier than in Botswana –

the ‘deserving poor’ in South Africa were seen to comprise individuals who

were too old, too infirm or too young to work. South Africa’s welfare system

was originally established to privilege white citizens, but over time had been

expanded to cover the black majority. The government led by the African

National Congress (ANC) therefore inherited in 1994 a social protection system

that provided generous non-contributory pensions to the elderly (costing 1.5

percent of GDP in 1994/95) and disabled (costing 0.5 percent of GDP), as well

as grants to some single mothers with children (costing 0.2 percent of GDP).

After 1994, the ANC government’s reforms were largely ‘parametric’, i.e. they

concerned the parameters of benefit levels and eligibility conditions of existing

programmes, rather than entirely new programmes (see Seekings & Nattrass,

2015: Chapter 6). The age of eligibility for old-age pensions was reduced to

sixty years for men, i.e. to the same level as for women. Access to disability

grants was first made easier, then tightened. Access to all pensions and grants

was extended to legally resident non-citizens. The level of benefits paid under

most programmes changed little in real terms (i.e. taking inflation into account).

Public employment programmes were expanded, and benefits increased.

Only one programme experienced dramatic reform: The democratic government

replaced the ‘state maintenance grant’ (SMG) it inherited from the apartheid

state with a new ‘child support grant’ (CSG), and subsequently raised the age

threshold at which children ceased to be eligible. This was a parametric reform

in the sense that it entailed changes to the level of the benefit and the conditions

21 This section is based on Seekings (2016d).

21

of eligibility. It was dramatic in the sense that it led to a massive increase in the

number of beneficiaries and a significant increase in real expenditure. In the

mid-1990s, the SMG was paid out for only about 200,000 children, at a cost of

about 0.2 percent of GDP. Almost none of these beneficiaries were black (or

African). In the mid-2010s, the CSG was paid out for about 12 million children,

at a cost of about 1.3 percent of GDP, making it one of a small number of social

assistance programmes in the global South that cost more than 1 percent of

GDP.

The government of South Africa never decided to spend an additional 1 or more

percent of GDP on this or any other social assistance programme. Rather, the

process of introducing and later expanding the CSG entailed at least six distinct

episodes of reform. In 1995-96, a government-appointed commission was tasked

with recommending whether to abolish the existing SMG, to expand access to

all South Africans (which would have cost 2 percent of GDP), or to reform it,

scaling back benefits whilst expanding access with the result that the overall cost

remained unchanged (at about 0.2 percent of GDP). The Commission

recommended the third option. During deliberations over the recommendation,

however, both benefits and access were improved somewhat, pushing the cost

up by an additional 0.2 percent of GDP. In 2002-03, another official commission

recommended that the means-test be abolished and/or the age threshold raised,

which would have cost an additional 1.6 percent of GDP. The government

rejected most of this proposal, instead raising the age threshold modestly at an

additional cost of only 0.4 percent of GDP. In 2005-08, the age-limit was again

discussed. In the face of proposals that would have cost an additional 0.4 percent

of GDP, the government again proceeded cautiously, raising the age threshold

only enough to cost an additional 0.1 percent of GDP. In 2008-09, the

government reformed the means-test (at a cost of 0.1 percent of GDP) and then

raised the age threshold once more, at a cost of 0.4 percent of GDP.

Political pressures pushed the government to expand this programme

incrementally. ‘Affordability’ was a constraint at times, but not at others.

Overall, its effect remains unclear. The initial cautious recommendation (in

1995-96) and the modest revision of benefits and age threshold (in 1997-98)

occurred against the backdrop of fiscal crisis, as the budget deficit and

government debt spiraled out of control. In these episodes, the fiscal context

appears to have been a constraint on expansion. The following two episodes of

reform – i.e. the expansions of 2002-03 and 2005-08 – occurred against the

backdrop of strong public finances, with revenues rising rapidly, the deficit

shifting to surplus, and government debt declining in relation to GDP. Yet these

expansions were only a little more generous than the reforms in the preceding

period of austerity. Healthy public finances in the 2000s did not lead to

profligacy; expensive reforms were rejected. The two final episodes of reform –

22

i.e. the expansions of 2008-09 – occurred against the backdrop of growing

budget deficits but still low levels of government debt. Despite the partial

deterioration in public finances, the expansions of 2008-09 cost in total as much

as the expansions in the preceding period of fiscal boom. There is no clear

correlation between fiscal space and policy reform.

The pattern of steady but non-linear expansion reflected political conditions.

The government could reject bold and expensive reform proposals because it

faced little external pressure, and the dominant ideology within the ANC was

skeptical about ‘handouts’ to working-age adults, even if the adults in question

were young mothers or other caregivers who were caring for children.22 At the

same time, the CSG had political champions within the ANC (especially a wily

Minister of Social Development), and cautious expansion made political sense

in the context of enduring poverty and rising inequality that threatened to

compromise electorally the ANC.

Zambia23

Zambia was the first country in East or central Africa to experiment with a cash

transfer programme but ten years later donor-funded pilot programmes had not

been scaled up significantly. Up to 2011, the government resisted cash transfers

in part on the grounds of unaffordability. A change of government in 2011

resulted in more support for policy reforms, but only limited reforms were

actually implemented. Given the pressures on public finances in Zambia,

affordability was a real concern. But, especially until 2011, it also masked a

deeper ideological hostility to ‘handouts’.

The need for cash transfers in Zambia arose from the combination of drought

(which intermittently devastated peasant agriculture) and AIDS (which resulted

in rising numbers of ‘labour-constrained’ households comprising the elderly,

sick people and children). In 2000, the government half-heartedly sought to

resuscitate its Public Welfare Assistance Scheme, i.e. poor relief dating back to

the colonial era. This came to naught, largely because neither the government

nor donors was willing to fund a system of ‘handouts’. In the early 2000s,

however, severe drought pushed donors into a massive emergency food relief

operation. Some of the donors decided that they should address the underlying

vulnerability that made people susceptible to drought rather than provide endless 22 One reform that was unambiguously not parametric was proposed (including by a

government-appointed commission of inquiry) but rejected (by the government): A basic

income grant, that would have been payable to all adults and children, including especially

adults of working age. 23 The first part of this section is based on Kabandula and Seekings (2014).

23

food aid. In 2002, the German Gesellschaft für Technische Zusammenarbeit

(GTZ) funded a ‘Social Safety Net Project’ in the Ministry of Community

Development an Social Services (MCDDSS), funded a study in 2003 of how it

might support people who had been impoverished by AIDS, and then (in late

2003) launched a pilot cash transfer scheme focused on ‘incapacitated’ – i.e.

‘labour-scarce and destitute’ – households. By early 2004 the scheme was

paying monthly benefits to just over one thousand households.

The Kalomo scheme led to further pilot schemes, mostly funded by DfID). The

purpose of the pilot schemes was primarily to generate the data needed to

convince the Zambian government that cash transfers were possible and

effective and would not have negative social effects. In other words, the pilot

schemes were essentially an exercise in political advocacy. But they were an

exercise that largely failed. From the outset it was clear that the donor-driven

pilot programmes enjoyed limited support from the government. Even within the

Department of Social Welfare (within the MCDSS), officials seem to have

favoured food aid programmes over cash transfers. The Ministry of Finance

blocked funds from the African Development Bank for a proposed child grant

programme, and declined to make any contribution from domestic revenues.

In the face of this deep ambivalence within the government, GTZ, DfID and

other donors sought to collect evidence of the efficacy of their experimental cash

transfer programmes. The Kalomo and later pilot programmes were subjected to

more-and-more thorough monitoring and evaluation (Van Ufford et al., 2016).

Donors pushed successfully for the establishment of a ‘sectoral advisory group’

for social protection, chaired by an MCDSS official, and bringing together

government and donor/agency personnel. The government incorporated a

rhetorical commitment to social protection in its Fifth National Development

Plan, published in December 2005. Opening an African Union (AU) conference

on social protection in Livingstone in 2006, Zambian president Levy

Mwanawasa described social protection as a ‘basic human right’ and as

affordable. But Mwanawasa’s own Minister of Finance, Ng’andu Magande,

conspicuously stayed away from the AU’s Livingstone conference, and the

Ministry of Finance continued to resist calls to scale up the pilot schemes. In

practice, the government continued to prioritise infrastructural and other

obviously developmental programmes.

In 2005, the GTZ’s consultant estimated that extending the Kalomo scheme “to

all of the 200,000 destitute households in Zambia” would cost about US$ 21

million, which was “the equivalent of 5% of the annual foreign aid inflow, or

0.5% of the Zambian GDP”. This was interpreted as showing that a countrywide

scheme was affordable (Schubert, 2005x: 24). MCDSS bureaucrats were flown

to a World Bank workshop on ‘mainstreaming’ social protection within poverty

24

reduction programmes. Growing support within the MCDSS resulted, in 2006-

07, in its adoption of an “implementation Framework for Scaling up a National

System on Cash Transfer”, providing for a ‘road map’ for rolling out social cash

transfers to more and more districts, culminating in complete country-wide

coverage by the end of 2012 (Chiwele, 2010: 48).

But the government remained unconvinced. A “nationally agreed

implementation plan” was needed before any roll-out could begin (quoted in

Chiwele, 2010: 6), and an implementation plan could not be agreed until there

was more evidence of the efficacy and affordability of cash transfers. DfID and

GTZ funded more research, as well as study tours to South Africa and Lesotho,

and radio and television programmes within Zambia. In a report for DfID,

Hickey et al. concluded that

‘the current regime cannot be described as particularly pro-poor and

the current Minister of Finance has repeatedly stated that poverty does

not exist, that “poor people” are simply lazy, and that policy should

focus on wealth creation than poverty reduction. This fits the

prevailing political discourse that tends to favour the “productive”

segment of the population’ (2009: 21).

Cash transfers might “have received favourable mentions in Presidential

speeches”, and proponents suggested “that the only obstacle to scaling-up the ST

pilots is now financial”, but “ownership within MCDSS is patchy and the

Ministry lacks the capacity and political clout to make serious headway with this

agenda”. Crucially, “key decision-makers in MFNP” were not convinced “of the

viability and desirability” of cash transfers; “there is little evidence that the

Minister of Finance and others in powerful positions have overcome their strong

ideological opposition” (ibid: 22). The following year, Chiwele assessed that

social protection had attracted “little political support” and the Ministry of

Finance remained “unconvinced regarding its economic merits”:

‘That MCDSS has problems rallying technocratic and political support

around a coherent agenda of social protection is also seen in the fact

that a Social Welfare Policy has been under development for the past

four years. The Ministry of Finance is still unconvinced. Other line

ministries continue to protect their areas and are not collaborating

effectively with the MCDSS. Social protection programmes are too

thinly scattered in various ministries. Larger ministries such as

Finance, Health, and Education are able to marshal a lot of support.’

(Chiwele, 2010: 5, 26).

25

Meanwhile, evidence accumulated on both the benefits and costs of cash transfer

programmes. A 2008 study by the International Labour Organisation (ILO),

funded by DfID, provided much fuller costing estimates than hitherto. The ILO

reported that public education (fully funded by the Zambian government) and

public health care (more than half of which was funded by donors) cost

approximately 14 percent of GDP, and this was predicted to rise slightly.

Nonetheless, the ILO concluded, there was ‘fiscal space’ to expand social

assistance programmes from their current meagre 0.2 percent of GDP (part of

which was funded externally, including by the WFP) to provide for ‘a minimum

social protection package’. This was defined as comprising ‘affordable universal

access to essential health care services; targeted social assistance; basic cash and

in-kind benefits for children (mothers and carers); and a basic universal pension

for the elderly and for persons with disabilities’ (ILO, 2008b: 5). First, the ILO

costed a national rollout of Kalomo-style cash transfers, targeted on the poorest

10 percent of households in each district; the cost of a benefit of just over K

50,000 (i.e. about $12) per household per month (rising with inflation) would

rise steadily to about 0.25 percent of GDP, in 2012. The study proposed that the

Zambian government assume full financial responsibility from 2012 (ibid: 106-

11). A universal old-age pension (from age 60, paying K 60,000 or about $15

per month) and child benefit schemes (paying lower benefits, for the first child)

at about 0.5 percent of GDP and 1.2 percent of GDP respectively (ibid: 164).

In 2011, a follow-up study by the ILO concluded that fiscal constraints meant

that schemes would need to be phased in over five years even with donors

meeting most of the cost at the outset and half of the cost after five years. Donor

funding would amount to almost 1 percent of GDP p.a., but this was a small

proportion of the total European Union aid to Zambia at the time. The study

assessed that the Zambian government could gradually increase its share of the

cost through additional taxation (Aguzzoni, 2011: xiv).

As the ILO recognised, the Zambian state did face fiscal constraints. Debt relief

in the early 2000s improved its external debt position, but it remained dependent

on grants to keep its budget deficit under control. The government’s response

was to contain expenditure. With steady growth of GDP, government

expenditure fell from more than 30 percent of GDP in the early 2000s to about

25 percent in the mid-2000s (ILO, 2008b). The result was that the government

had more fiscal space than hitherto, but was reluctant to increase expenditure for

fear of returning to the fiscal problems of earlier years. The government did,

however, fund heavily agricultural programmes, including means-tested

programmes intended to assist poor farmers to increase production.

In practice, the government of Zambia continued to underfund social protection.

The budget allocation to social protection remained less than 3 percent in 2009,

26

2010 and 2011. The Sixth NDP, published in January 2011, referred grandly to

the expansion of cash transfers to 300,000 ‘incapacitated’ households by 2015

(as well as of other cash transfer and feeding programmes). But the budgetary

allocations did not match this scale-up. The projected allocation to social

assistance barely rose between 2011 and 2015, despite the envisaged explosion

in the number of beneficiaries (Zambia, 2011). Interviews suggest that the

chapter on social protection in the Sixth NDP was dropped entirely from the

penultimate draft, before being re-included after strident protests from civil

society and donors.24 There remained an evident lack of political will on the part

of the government.

In the 2011 elections, however, the incumbent Movement for Multiparty

Democracy (MMD) lost to the populist Patriotic Front, led by Michael Sata. A

social democratic faction within the MMD had ensured the inclusion of a

commitment to expanding social protection in the party’s 2011 election

manifesto. Following his election, Sata restated this commitment, but the

number of households receiving social cash transfers only slowly increased in

line with the Sixth NDP. It was not until October 2013 that the Finance Minister

actually announced an increase in the government’s budget allocation for cash

transfers beyond the modest growth envisaged by the previous government. The

increased contribution sounded large – a ‘700 percent increase’ – but this was

from a tiny base, with only 61,000 households receiving cash transfers, with

three-quarters of the costs covered by donors. The following year the PF

government published its Revised Sixth NDP, which included a promise to

scale-up the poverty-targeted ‘social cash transfer’ scheme to 500,000

households by 2016. By 2015 the number of beneficiary households had tripled,

to 190,000 households, with donors paying for only one-sixth of the total cost

(less in current prices than in 2013) (Siachiwena, 2016).

The 2013-14 reforms represented an important shift, but even the new plans

were very modest in comparison with the proposals made by the ILO. The

budgeted expenditure in 2015 amounted to only 0.1 percent of GDP,25 compared

to the approximately 2 percent of GDP in the ILO proposals.

For about ten years, from 2003 to 2013, international donors and agencies tried

but failed to persuade the government of Zambia to implement its promised

scale up of cash transfer programme or to commit necessary funds. Only in

2013-14 did the government commit any funds, and then the cost amounted to

only 0.1 percent of GDP. Donors and agencies made relentless efforts to collect

and present evidence showing that a set of programmes was affordable, poverty-

24 Interview with Mutale Wakunuma, 20 March, 2014. 25 Budgeted expenditure of ZMK 180 million; GDP in current prices of ZMK 190 billion.

27

reducing and developmental, contributing to economic growth. But they failed

to persuade the MMD government in office until 2011, and had limited success

during the subsequent Sata government.

For the first five years – from 2003 to 2008 – opposition was personified by the

Minister of Finance and National Planning, Nga’ndu Magande. Magande

fervently opposed cash transfers on largely neoliberal grounds. In his view,

laziness was the primary cause of poverty, and ‘handouts’ promoted laziness and

a ‘culture of dependency’. Recipients, Magande claimed, spent benefits on

alcohol. He was opposed even to old-age pensions, because these discouraged

people from saving for their old age. Magande was indifferent to donors who

chose to fund cash transfer programmes, but the Zambian government should

not allocate scarce resources – or accumulate debt – in order to fund

unproductive ‘handouts’ (Kabandula & Seekings, 2014).

Magande’s views may be extreme, but they were not entirely out-of-line with

the ideology of other personal MMD leaders, which explains why there was

little or no pressure from other senior party members in government to scale up.

Not even the Minister of Community Development supported cash transfers

enthusiastically.26 The MMD had been formed in reaction to Kaunda regime.

Kaunda had abolished multi-party democracy, so the MMD championed it.

Kaunda had adopted disastrous statist economic policies, so the MMD embraced

free markets. In the introduction to its 1991 manifesto, the MMD had declared:

‘MMD believes that economic prosperity for all can best be created by

free men and women through free enterprise; by economic and social

justice involving all the productive resources – human, material and

financial, and by liberalising the industry, trade and commerce, with

the government only creating an enabling environment whereby

economic growth must follow as it has in all the world’s successful

countries’ (MMD, 1991).

The 1991 manifesto – and subsequent party documents – referred to ‘safety nets’

for the ‘destitute’, but it was clear that the government’s responsibility was

minimally residual. From 2007, in the face of stronger challenge from Sata and

the PF, the MMD made more generous promises, but its priority remained

clearly increased production through free enterprise and a smaller state (Larmer

& Fraser, 2007; Cheeseman & Hinfelaar, 2010).

Sata and the PF government were more populist in general (Resnick, 2014), and

a faction within the PF advocated expanded social protection. But even after

26 Interview, Michael Kaingu, 10th March 2014.

28

2013 the PF’s plans remained very modest, with expenditure rising to only 0.1

percent of GDP. The PF might have emphasized its commitment to “delivering

inclusive development and social justice” (which was the theme of its 2013

budget that made provision for increased expenditure), and it might have said

that it sought to reallocate funds to social protection from the MMD’s subsidy

programmes. In practice, however, the PF retained its predecessor’s productivist

priorities and skepticism about cash transfers. As the Finance Minister made

clear in his 2012 budget speech, inclusive development and social justice “will

not be achieved simply by promoting handouts”; rather he explained, the PF’s

“fundamental approach is to build a self-reliant people able to sustainably

generate money for their own pockets”.27

In sum, successive MMD and PF governments in Zambia favoured other areas

of public expenditure – including especially support for farmers – over social

protection. The government did face fiscal constraints, but these cannot explain

the allocation of only 0.1 percent of GDP to the cash transfer programmes.

Pension reform in Zanzibar28

In Zanzibar, which comprises a semi-autonomous territory within the United

Republic of Tanzania, non-contributory, tax-financed old-age pensions were

introduced in April 2016. The programme was universal – meaning that there

was no means-test – but was limited to men and women from the age of seventy.

The value of the pension was a modest TZS 20,000 or just under US$ 10 per

month. It was funded fully through the budget of the Government of Zanzibar,

without any direct funding by foreign aid donors. Zanzibar’s recently re-elected

President, Dr Ali Mohamed Shein, expressed his strong support for the new

programme and for gradual increases in the value of the pension: “If we record

admirable growth of our economy, definitely the amount for universal social

pension will be increased. Our elderly should accept our donation and be patient,

looking forward for better future”.

The government of Zanzibar declared repeatedly its commitment to reducing

poverty. It saw economic growth as the most important mechanism for doing so,

but acknowledged its responsibility for “the poor majority in society”, including

through ‘safety nets’. The government had long administered a system of poor

relief through discretionary posho (i.e. small payments or allowances) and

operated residential homes for some elderly men and women. Up until the

2010s, however, the governing elite seemed to prefer religious charity to

27 Budget speech, October 2012. 28 This section is based on Seekings (2016e).

29

government safety net programmes. As across much of Africa, conservatism

within government was challenged from the outside, by international agencies,

donors and NGOs. One powerful influence was the World Bank, which helped

to push the government into a conditional cash transfer programme for poor

families with children, largely funded by the World Bank. This programme –

which formed part of the third phase of the Tanzanian Social Action Fund –

provided between TZS 10,000 and 30,000 per month (depending on the number

of children in the household) to a total of about 33,000 households by September

2016. A competing set of agencies – including primarily the ILO and HelpAge

International (henceforth simply HelpAge) – pushed energetically for a

universal, non-contributory old-age pension. In 2009, in a report written with the

Ministry of Health and Social Welfare, HelpAge recommended a pension

costing 0.85 percent of GDP (with other, more expensive options costing up to

double this). In January 2010, the ILO (in collaboration with the Ministry of

Labour, Youth, Women and Children Development) costed a pension for men

and women from the age of 60, paying TZS 15,000 per month (in 2009 prices),

at approximately 1.2 percent of GDP.

Support for the proposed pension within the state was initially concentrated

within the Department of Social Welfare. In 2011-12, the Department began to

prepare an overarching ‘Social Protection Policy’, established a Social

Protection Unit, and sent personnel off on relevant courses. Working closely

with HelpAge, the new Ministry began to sell the idea of a pension to

bureaucrats in other government departments, government ministers and

Members of Parliament. In early 2013, HelpAge organized a study tour of

Mauritius by a team of government officials and politicians. The Social

Protection Policy was drafted, including an explicit commitment to universal

pensions, presented as an extension of the ‘existing social pension scheme’ (i.e.

posho).

These efforts helped to build support for a pension, but there remained enough

opposition in other ministries to stall the process. When the draft Social

Protection Policy was presented to the Cabinet in early 2014, some ministers

queried the proposed pension on the grounds of its affordability. The President

himself was reportedly in favour, but he proposed that the issue should be

referred to a technical committee to examine whether – or what kind of – a

pension was affordable. Whilst the President’s motivation and intention were

unclear, the decision made it more likely that the government’s final decision

would be broadly consensual and would appear technocratic. The committee (or

‘task team’) comprised a mix of pro-reform and skeptical bureaucrats. The

committee’s first meeting was difficult: “The issue was the budget, what could

we pay?”, recalls one member of the Task Team; “the Ministry of Finance was

reluctant to spend money, they said this is what we can afford”. “We all knew

30

that the Ministry of Finance had no money”, recalls another member of the Task

Team, but even proposals for a modest programme provoked opposition.

Echoing the World Bank, skeptics argued that committing funds for the pension

would deprive other programmes, for example infrastructural investments or

programmes for the youth. They worried that Zanzibar’s foreign donors (“our

development partners”) don’t like profligacy. They contrasted the proposed

pension with the World Bank-funded conditional cash transfers, which were

represented as “helping people to use their talent, to use their energy; it is not

paying people to sit down and do nothing”. The committee made progress only

when the sceptics were assured that their concerns were being taken seriously.

The Government’s financial position was indeed parlous, despite strong

economic growth. The following May the Minister of Finance summarized the

state of Zanzibar’s economy and public finances in his budget speech. Despite

healthy economic growth, the government remained heavily dependent on

foreign aid, which funded most of the development budget and about one-tenth

of recurrent expenditure (through general budget support). The Government was

expecting to raise TZS 375 billion in the 2014-15 year, with foreign aid

amounting to more than TZS 300 billion. The proposed pension would entail a

sizeable addition to the recurrent expenditure budget.

The question of affordability depended on more specific choices: Should a

pension be universal or targeted? At what level should the pension be set? From

what age should pensions be paid? The skeptics quickly conceded that the

pensions should be universal and agreed to set benefits at about TZS 20,000 per

month. The stumbling block was the age threshold. Social Welfare officials soon

abandoned the goal of paying pensions from the age of 60, as recommended in

the reports from HelpAge and ILO in 2009-10. The Task Team’s report

recommended an age threshold of 65. This would cost TZS 9.6 billion p.a. or 0.7

percent of GDP. The Task Team added, however, that if resources did not

permit this, then the threshold should be 70, which would cost only TZS 6.6

billion p.a. or 0.5 percent of GDP. Proponents of a younger age threshold

arranged for the former chief bureaucrat in the Mauritian Ministry of Social

Security and National Solidarity to visit Zanzibar to meet with key government

officials and ministers in order to persuade them to follow the lead of Mauritius

in implementing a universal pension. Despite this, however, the Cabinet opted in

March 2015 to introduce pensions only from the age of 70. This meant that,

when payments began in 2016, only about 25,000 people received them, and the

total cost was only about 0.4 percent of GDP, which was far below the options

put forward by HelpAge and ILO in 2009-10.

In several respects, the broader social, economic and political context in

Zanzibar was favourable to the introduction of a pension. Firstly, there was a

31

clear need for financial assistance for the elderly. To a greater extent than on the

mainland, the Zanzibari economy and society were undergoing ‘de-

agrarianisation’ in that the roles of subsistence agriculture and kin-based

redistribution and care were both declining. Deep poverty persisted despite

economic growth, as the benefits of growth barely trickled down to the very

poor. Secondly, Zanzibar had a tradition of public responsibility for the poor, so

that advocates of the pension could represent their proposals as improvements

rather than entirely novel departures from existing policies. Moreover, at no

point did skeptics or opponents invoke the discourse of ‘handouts’ and

‘dependency’ that is widespread elsewhere in East Africa. In Zanzibar, in

contrast, the hegemonic public discourse was one of responsibility, not

dependency. Thirdly, Zanzibar had very competitive elections, giving the

incumbent president and party strong electoral incentives to introduce a popular

reform. Both President Karume (in 2009) and his successor President Shain

seem to have favoured reform. The fiscal context was not favourable, however.

The Minister of Finance was broadly supportive, but there was opposition – or,

at least, ambivalence – among senior bureaucrats in various ministries. The

cabinet’s decision in 2015 to introduce a universal pension but with the high age

threshold of 70 represented a compromise.

The Zanzibari case suggests that concerns about affordability need to be

addressed directly rather than side-stepped. Advocates of welfare expansion like

to invoke the rights stipulated in international conventions and declarations. But

fiscal anxieties are unlikely to be dispelled by any rhetoric of rights. In Zanzibar,

fiscal conservatives felt that their concerns were taken seriously, that different

options were costed properly, and both the benefits and costs of each option

were made clear to the Cabinet, which could make an informed decision.

Conclusion: The politics of (un)affordability

A government’s assertion that a policy reform is unaffordable might reflect any

one of several underlying arguments. First, it might reflect an assessment that

the economic costs of raising additional revenues, whether through tax or debt,

outweigh the benefits of the reform. This is especially likely if, in the

government’s view, the additional government expenditure inhibits rather than

promotes economic growth. Alternatively, it might reflect an assessment that the

political costs of additional taxation outweigh the political benefits of

redistributing resources to some poor citizens. Thirdly, it might reflect the

assessment that this particular reform is not a priority, and scarce funds should

be spent on other programmes. Other programmes might be more conducive of

economic growth, or might provide the incumbent president or party with more

32

useful political rewards, or might simply reflect a normative or ideological

preference.

In exceptional circumstances, the economic costs of increased government

spending are minimal. Such circumstances existed in Botswana during some of

the 1980s and early 1990s, when rapidly rising mining revenues encouraged

profligacy. Since the late 1990s, however, the governments in all four countries

considered here were well aware of fiscal constraints, at least in the sense that

the opportunity cost of increased expenditure on one programme was less new

funding for other programmes. Governments in all four countries have limited

reforms so as to contain expenditure. Even in Botswana, in 1996, the

government set the value of the new old-age pension at a low level (most

obviously in comparison with neighbouring South Africa) with the effect that

the cost of the programme was only about 0.5 percent of GDP.

In these four country case-studies, the most expensive reforms undertaken since

the early 1990s each cost 0.4 to 0.5 percent of GDP. These included, in

Botswana, the government’s assumption of financial responsibility for WFP

feeding programmes and its introduction of old-age pensions. The Zanzibar

government’s introduction of old-age pensions in 2015-16 was predicted to cost

0.5 percent of GDP, although by the time the first pensions were paid the cost

was closer to 0.4 percent of GDP. The South African government’s decisions to

expand its Child Support Grant each cost about 0.4 percent of GDP. All four

governments rejected proposals for reforms that would cost more than this. In

South Africa, child grant reforms that would have cost 1.6 to 2 percent of GDP

were rejected summarily in the late 1990s and early 2000s. In Botswana, the

government decided against setting pension benefits at a more generous level

that would have raised the programme’s cost above 0.5 percent of GDP, and

later rejected proposed Child and Family Support Grants. In Zanzibar, the

government decided against setting the age threshold lower than 70, which

would have raised the programme’s cost above 0.5 percent of GDP. In Zambia,

there is no evidence that either the MMD or PF governments considered

seriously proposals from the ILO for pensions that would have cost about 0.5

percent of GDP or for child grants that would have cost more than 1 percent of

GDP.

Many of the reforms cost a lot less than this. In Zambia, most obviously, even

the post-2013 reforms by the PF government cost only 0.1 percent of GDP. In

Botswana, in 2012/13, the budget for the old-age pension programme had fallen

to 0.2 percent of GDP as the economy had grown and benefits remained modest.

The Orphan Care and Destitute programmes each cost about 0.2 percent of

GDP. Feeding programmes cost more, in total, at about 0.5 percent of GDP,

whilst the Ipelegeng public works programme cost about 0.3 percent of GDP

33

(BIDPA & World Bank, 2013: x). Whilst the total cost of these safety net

programmes in Botswana came to about 1.7 percent of GDP, the individual

programmes were low-cost. In South Africa, only the Child Support Grant was

expanded substantially, with the cost of other programmes changing little over

time, as the real value of benefits was not increased and growth in the number of

beneficiaries was contained.

It seems that reforms costing more than 0.4 to 0.5 percent of GDP are not

politically feasible in countries such as the four cases considered in this paper.

Often, reforms costing a lot less than this are not feasible. Precisely why is not

the subject of this paper, but a large part of the answer involves widespread

ideological or normative opposition to ‘handouts’ among African political elites

(see also Kalebe-Nyamongo & Marquette, 2014, on Malawi), such that even in

competitive party systems incumbent parties are reluctant to promise and then

implement bold reforms. The point for this paper is that the constraint on

programmatic reform does not appear to be affordability in simple economic or

fiscal terms. The political ceiling on reforms is far lower than the ceiling

suggested in technical studies of fiscal space. In practice, ‘affordability’ in

Africa is not about fiscal space, but is a matter of political priorities.

34

References

Aguzzoni, L. 2011. The concept of fiscal space and its applicability to the

development of social protection policy in Zambia. ESS Paper, 28. Geneva:

International Labour Organisation, Extension of Social Security.

BIDPA & World Bank. 2013. Botswana Social Protection Assessment.

Gaborone: Botswana Institute for Development Policy Analysis and Washington

DC: World Bank.

Botswana. 1997. National Development Plan 8, 1997/98-2002/03. Gaborone:

Ministry of Finance and Development Planning.

-----. 2003. National Development Plan 9, 2003/04-2008/09. Gaborone: Ministry

of Finance and Development Planning.

-----. 2009. National Development Plan 10, 2009-2016. Gaborone: Ministry of

Finance and Development Planning.

-----. 2013. Mid-term review of NDP 10. Gaborone: Ministry of Finance and

Development Planning.

Cheeseman, N. & M. Hinfelaar. 2010. Parties, Platforms, and Political

Mobilization: The Zambian Presidential Election of 2008. Journal of African

Affairs, 109(434): 51-76.

Chiwele, K.D. 2010. Assessing the Administrative Capacity and Costs of Cash

Transfer Schemes in Zambia. Brasilia: International Policy Centre for Inclusive

Growth.

Colclough, C. & S. McCarthy. 1980. The Political Economy of Botswana: A

Study of Growth and Distribution. Oxford: Oxford University Press.

Garcia, M. & C.M.T. Moore. 2012. The Cash Dividend: The Rise of Cash

Transfer Programs in Sub-Saharan Africa. Washington D.C.: World Bank.

Hagen-Zanker, J. & A. McCord. 2010. Financing Social Protection in the Light

of International Spending Targets: A Public Sector Spending Review. London:

Overseas Development Institute.

Handley, G. 2009. Fiscal space for strengthened social protection: West and

Central Africa. London: Overseas Development Institute and UNICEF.

HelpAge International. 2009. Social Protection Policy: Responses to Older

People’s Needs in Zanzibar. London: HelpAge International.

Hickey, S., Wheeler, R.S., Guenther, G. & I. Macauslan. 2009. Promoting SP &

ST: DFID the Politics of Influencing. Working Paper, 32. London: Department

for International Development.

Holm, J. & R.G. Morgan. 1985. Coping with drought in Botswana: an African

success. Journal of Modern African Studies, 23(3): 463-482.

ILO. 2006. Cash benefit in low-income countries: Simulating the effects on

poverty reduction for Senegal and Tanzania. Issues in Social Protection

discussion paper, 15. Geneva: International Labour Organisation, Social

Security Department.

-----. 2008a. Can Low-income Countries Afford Basic Social Security? Social

Security Policy Briefings, 3. Geneva: International Labour Organization.

-----. 2008b. Zambia: Social Protection Expenditure and Performance Review

and Social Budget: Executive Summary. Geneva: International Labour

Organisation.

-----. 2008c. Tanzania Mainland: Social Protection Expenditure and

Performance Review and Social Budget, Executive Summary. Geneva:

International Labour Organisation.

-----. 2010. Zanzibar: Social Protection Expenditure and Performance Review

and Social Budget. Geneva: International Labour Organisation.

-----. 2014. World Social Protection Report 2014/15: Building Economic

Recovery, Inclusive Development and Social Justice. Geneva: International

Labour Organisation.

-----. 2015a. Social Protection for Children: Key Policy Trends and Statistics.

Social Protection Policy Paper, 14. Geneva: International Labour Organisation,

Social Protection Department.

-----. 2015b. Fiscal Space for Social Protection: Option to Expand Social

Investments in 187 Countries. Extension of Social Security Working Paper, 48.

Geneva: International Labour Organisation.

Kabandula, A. & J. Seekings. 2014. Donor influence, the Minister of Finance

and the delayed reform of welfare policy in Zambia, 2003-11. Paper presented at

workshop on Social Protection in Africa, University of Cape Town.

Kakwani, N. & K. Subbarao. 2005. Aging and Poverty in Africa and the Role of

Social Pensions. Washington D.C.: World Bank.

----- & ----. 2007. Poverty among the elderly in Sub-Saharan Africa and the role

of social pensions. Journal of Development Studies, 43(6): 987-1008.

Kjaer, A.M & M.S. Ulriksen. 2014. From Consensus to Contention: Changing

Revenue and Policy Dynamics in Uganda. Working Paper, 2014-21. Geneva:

UNRISD.

Larmer, M. & A. Fraser. 2007. Of cabbages and King Cobra: Populist politics

and Zambia's 2006 election. African Affairs, 106(425): 611-637.

Magang, D. 2008. The Magic of Perseverance: The Autobiography of David

Magang. Cape Town: Centre for Advanced Studies of African Society.

MMD. 1991. Election Manifesto 1991. Lusaka: Movement for Multiparty

Democracy.

O’Cleirigh, E. 2009. Affordability of Social Protection Measures in Poor

Developing Countries. Promoting Pro-Poor Growth: Social Protection, 111-

127.. Paris: OECD.

Pal, K., Behrendt, C., Leger, F., Cichon, M. & K. Hagemejer. 2005. Can Low-

income Countries Afford Basic Social Security? First Results of a Modelling

Exercise Issues in Social Protection. Discussion Paper, 13. Geneva,

International Labour Office.

Ravallion, M. 2008. Bailing out the World’s Poorest. Policy Research Working

Paper, 4763. Washington D.C.: World Bank.

REPOA. 2010. Social Protection of the Elderly in Tanzania: Current Status and

Future Possibilities. REPOA Special Paper, 10/5. Dar es Salaam: Research on

Poverty Alleviation.

Resnick, D. 2014. Urban Poverty and Party Populism in African Democracies.

Cambridge: Cambridge University Press.

Samson, M. & C. Cherrier. 2009. Feasibility Study for a Social Cash Transfer

Programme as a Lead Instrument in Child-Centered Social Protection in

Senegal. Cape Town: Economic Policy Research Institute.

Schnitzer, P. 2011. Design Options, Costs and Impacts of a Cash Transfer

Program in Senegal: Simulation Results. Dakar: UNICEF.

Schubert, B. 2005. Social cash transfers – reaching the poorest: A contribution

to the international debate based on the experience of Zambia. Eschborn: GTZ.

Seekings, J. 2013. Welfare regimes and distribution across the global South:

Theory and evidence in the construction of typologies. Unpublished paper.

-----. 2014. Are African welfare states different? Welfare state-building in

(Anglophone) Africa in comparative perspective. Paper presented at workshop

on Social policy and regimes of social welfare in Africa. Fribourg.

-----. 2016a. Building a conservative welfare state in Botswana. CSSR Working

Paper No 389. Cape Town: Centre for Social Science Research, University of

Cape Town.

-----. 2016b. Drought relief and the origins of a conservative welfare state in

Botswana, 1965-1980. CSSR Working Paper No. 378. Cape Town: Centre for

Social Science Research, University of Cape Town.

-----. 2016c. “A lean cow cannot climb out of the mud, but a good cattleman

does not leave it to perish”: The origins of a conservative welfare doctrine in

Botswana under Seretse Khama, 1966-1980. CSSR Working Paper No. 388.

Cape Town: Centre for Social Science Research, University of Cape Town.

-----. 2016d. Redefining the “affordability” of social assistance programmes:

The Child Support Grant in South Africa, 1998-2014. CSSR Working Paper No.

379. Cape Town: Centre for Social Science Research, University of Cape Town.

-----. 2016e. The Introduction of Universal Old Age Pensions in Zanzibar. CSSR

Working Paper No. 393. Cape Town: Centre for Social Science Research,

University of Cape Town.

----- & N. Nattrass. 2015. Policy, Politics and Poverty in South Africa. London:

Palgrave.

Siachiwena, H. 2016. Social protection policy reform in Zambia during the Sata

presidency, 2011-2014. CSSR Working Paper No. 380. Cape Town: Centre for

Social Science Research, University of Cape Town.

Stevens, C. 1978. Food aid and nutrition: The case of Botswana. Food Policy:

18-28.

UNECA. 2011. The African Social Development Review. Addis Ababa: UN

Economic Commission for Africa, Human and Social Development Section.

Van Ufford, P.Q., Harland, C., Michelo, S., Tembo, G., Toole, K. & D. Wood.

2016. The Role of Impact Evaluation in the Evolution of Zambia’s Cash

Transfer Programme. In Davis, B. et al. (eds), From Evidence to Action: The

Story of Cash Transfers and Impact Evaluation in Sub-Saharan Africa, 197-225.

Oxford: Oxford University Press, FAO and UNICEF.

World Bank. 2012a. Affordability and Financing of Social Protection Systems.

Washington D.C.: Africa Social Protection Policy Brief.

-----. 2012b. Managing Risk, Promoting Growth: Developing Systems for Social

Protection in Africa. Washington D.C.: World Bank.

-----. 2015. The State of Social Safety Nets 2015. Washington D.C.: World Bank.

Wright, G. & M. Noble. 2010. Recent Social Policy Developments in Africa.

Global Social Policy, 10(1): 111-119.

Zambia. 2011. Sixth National Development Plan 2011-2015. Lusaka.